Talking Robots: Josh Bongard and Hod Lipson - Resilient Machines through continuous self modeling

Autonomous robots capable of self-modeling can maintain functionality despite physical damage by creating and updating internal simulations of their own morphology. By observing sensory feedback from their movements, these machines infer their current physical state without prior human-provided models, enabling them to adapt to structural failures like broken limbs. This research, led by Hod Lipson and Josh Bongard, demonstrates how robots can autonomously generate new gaits to compensate for injury, mirroring biological processes of self-awareness and curiosity. Beyond robotics, this methodology applies to structural health monitoring, where systems like bridges could detect and report internal degradation. Future advancements aim to scale this technology to multi-robot systems, allowing groups of machines to share experiences and collectively improve their resilience in complex, unstructured environments.